Carbon emission monitoring and predicting method and device based on digital twinning and medium

Abstract

The invention discloses a carbon emission monitoring and predicting method and device based on digital twinning and a medium, and relates to the technical field of park carbon emission monitoring. The method comprises the steps of collecting multi-source data of a target area in real time through a multi-source heterogeneous sensor; performing dynamic anomaly detection on the multi-source data, and triggering cross-data type relevance verification when data anomaly is detected, so as to select a repair strategy according to a verification result; dynamically binding the repaired data with the space coordinates of the BIM model, and constructing a three-dimensional digital twinborn body containing an equipment-level carbon flow path; performing rolling prediction on the carbon emission trend based on a preset hybrid prediction model, and executing a control instruction from carbon emission source positioning to equipment parameter adjustment step by step when a predicted value triggers a multi-stage early warning logic; and after the control instruction is executed, the control effect is verified through real-time feedback of the three-dimensional digital twinborn body, and root cause backtracking analysis is started when an expectation is not reached.

Description

Technical Field

[0001] This application relates to the technical field of park carbon emission monitoring, and particularly to a carbon emission monitoring and prediction method, device, and medium based on digital twin. Background Art

[0002] With the in-depth promotion of the "dual carbon" goal, industrial parks, as key monitoring objects of carbon emissions, their precise monitoring and intelligent management have become key links in achieving low-carbon transformation. Traditional carbon emission monitoring technologies generally have systematic defects such as discrete data acquisition, lagged analysis, and empirical decision-making, making it difficult to meet the urgent needs of refined management in modern parks.

[0003] Existing technologies usually rely on the offline detection mode of manual sampling combined with laboratory analysis, with a monitoring cycle of up to several days, resulting in carbon emission data being seriously lagged behind the actual production process and unable to capture the instantaneous emission characteristics in the dynamic production process. At the same time, there are technical bottlenecks in the fusion processing of multi-source heterogeneous data. Data from energy management systems, environmental monitoring devices, and production control systems form multiple data islands due to protocol differences and time asynchronization, making it difficult to ensure the integrity and accuracy of carbon emission accounting.

[0004] In terms of prediction and early warning, traditional time series models have insufficient adaptability to complex working conditions. Especially when facing non-linear scenarios such as sudden changes in production schedules and equipment start-stop, the prediction error rate increases significantly, making it difficult to provide a reliable basis for emission reduction decisions. At the visualization level, existing two-dimensional display systems lack in-depth integration of spatial dimensions and are difficult to accurately map carbon emission hotspots to specific production equipment, resulting in low efficiency of traceability analysis.

[0005] More notably, most monitoring platforms have not yet built a closed-loop control system of "monitoring - prediction - regulation". Decision support still stays at the stage of manual experience judgment, lacking an intelligent decision-making mechanism based on multi-dimensional simulation and deduction, resulting in a significant reduction in the scientificity and timeliness of emission reduction plans. The above technical defects not only affect the real-time and accurate measurement of carbon emissions, but also restrict the optimal allocation of carbon assets and the effective implementation of emission reduction measures, and have become an important technical barrier restricting the green upgrading of industrial parks.

[0006] Therefore, there is an urgent need to develop a new type of carbon emission monitoring solution that integrates Internet of Things perception, digital twin modeling, and intelligent decision-making, and break through the limitations of existing methods through technological innovation to provide full-chain technical support for park-level carbon management. Summary of the Invention

[0007] In view of the above problems, this application provides a carbon emission monitoring and prediction method, device, and medium based on digital twin that overcomes the above problems or at least partially solves the above problems. The technical solutions are as follows: In a first aspect, an embodiment of the present application provides a carbon emission monitoring and prediction method based on digital twin. The method includes: real-time collecting multi-source data of a target area through multi-source heterogeneous sensors; where the multi-source data includes: energy data, material data, environmental data, equipment operation data, and spatial topology data; performing dynamic anomaly detection on the multi-source data, and triggering cross-data-type correlation verification when data anomalies are detected, so as to select a repair strategy according to the verification result; dynamically binding the repaired data with the spatial coordinates of the BIM model to construct a three-dimensional digital twin including equipment-level carbon flow paths; performing rolling prediction on the carbon emission trend based on a preset hybrid prediction model, and when the predicted value triggers a multi-level early warning logic, successively executing control instructions from carbon emission source location to equipment parameter adjustment; after the control instructions are executed, verifying the control effect through the real-time feedback of the three-dimensional digital twin, and initiating root cause backtracking analysis when the expectation is not met.

[0008] In an implementation manner of the present application, performing dynamic anomaly detection on the multi-source data, and triggering cross-data-type correlation verification when data anomalies are detected, so as to select a repair strategy according to the verification result, specifically includes: constructing a dynamic baseline model for the energy data, and judging the deviation magnitude value of each data point in the energy data from the preset baseline in the dynamic baseline model; if the deviation magnitude value exceeds the preset fluctuation range, marking the corresponding data point as a suspected anomaly, and simultaneously verifying the corresponding material conservation relationship in the material data; if the material conservation error is within the tolerance range, determining that the energy data sensor fails and triggering repair; if the material conservation error exceeds the tolerance range, determining it as a real anomaly event and triggering a multi-data stream joint repair algorithm.

[0009] In an implementation manner of the present application, after triggering the multi-data stream joint repair algorithm, the method further includes: extracting the temperature and humidity change characteristics in the environmental data during the abnormal period; constructing an equipment state transition matrix based on the start-stop records in the equipment operation log; solving an optimal repair value set that makes the energy data, material data, and environmental data compatible through a constraint satisfaction algorithm.

[0010] In an implementation manner of the present application, dynamically binding the repaired data with the spatial coordinates of the BIM model to construct a three-dimensional digital twin including equipment-level carbon flow paths, specifically includes: in the initial mapping stage, dynamically associating the HVAC pipeline ID in the BIM model with the real-time energy consumption data; when detecting carbon emission anomalies of a specific equipment, automatically loading the fluid dynamics simulation model of the specific equipment; overlaying and displaying the real-time sensor data and the simulation data, and characterizing the carbon emission intensity deviation through a color temperature gradient.

[0011] In an implementation manner of the present application, the multi-level early warning logic includes: primary early warning logic, intermediate early warning logic, and advanced early warning logic; the primary early warning logic is to mark potential risk devices in the digital twin when the predicted value reaches the first early warning threshold; the intermediate early warning logic is to generate an equipment operation parameter optimization plan and simulate the implementation effect when the predicted value reaches the second early warning threshold; the advanced early warning logic is to trigger the equipment interlock protection mechanism and freeze the high-risk process section when the predicted value exceeds the third early warning threshold.

[0012] In an implementation manner of the present application, generating an equipment operation parameter optimization plan specifically includes: constructing a carbon emission efficiency curve for each equipment based on the real-time monitoring data of the three-dimensional digital twin, where the carbon emission efficiency curve reflects the carbon emissions per unit output of the equipment at different load rates; using the carbon emission efficiency curve as an input parameter to construct a multi-objective optimization model with the goals of minimizing carbon emissions and maximizing production efficiency; calculating the Pareto optimal solution set through the multi-objective optimization model to obtain several candidate adjustment plans including the trade-off relationship between carbon emissions and production efficiency; inputting the candidate adjustment plans into the three-dimensional digital twin for virtual execution simulation to predict the impact of each candidate adjustment plan on the carbon emission trend and equipment operation stability; screening feasible plans that meet the preset control goals according to the simulation results, generating equipment parameter priority adjustment instructions and sending them to the equipment control system for execution; real-time monitoring the carbon emission data after the execution of the instructions and transmitting the feedback results back to the multi-objective optimization model for dynamic weight calibration.

[0013] In an implementation manner of the present application, root cause backtracking analysis specifically includes: intercepting multi-source data in each preset time period before and after the control instruction is issued; reconstructing the control instruction propagation path and equipment response time sequence in the three-dimensional digital twin; identifying the key nodes of control failure through a causal inference model; where the key nodes include sensor delay, model prediction deviation, or equipment actuator failure.

[0014] In an implementation manner of the present application, the method further includes: when the root cause backtracking analysis identifies a prediction model deviation, automatically extracting the feature vector of the failure case; injecting the feature vector into the adversarial training set to generate a new model version with perturbation robustness; running the old and new models in parallel in the three-dimensional digital twin, and selecting the optimal model to go online through real-time data verification.

[0015] In a second aspect, an embodiment of the present application further provides a carbon emission monitoring and prediction device based on digital twin. The device includes: at least one processor; and a memory communicatively connected to the at least one processor; where the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a carbon emission monitoring and prediction method based on digital twin as described in any one of the above.

[0016] In the third aspect, an embodiment of the present application also provides a non-volatile computer storage medium for carbon emission monitoring and prediction based on digital twins, which stores computer executable instructions. When the computer executable instructions are executed, a carbon emission monitoring and prediction method based on digital twins such as any of the above is implemented.

[0017] The carbon emission monitoring and prediction method, device and medium based on digital twin provided in the embodiments of the present application have the following beneficial effects: 1. This application has achieved a systematic innovation of the technical paradigm in the field of industrial carbon emission management by constructing a collaborative technology system of multi-source data fusion acquisition, dynamic digital twin mapping, hybrid intelligent prediction and closed-loop control feedback. Based on the real-time synchronous acquisition and cross-modal correlation verification mechanism of multi-source heterogeneous data, it effectively solves the monitoring blind spot problem caused by data islands in traditional monitoring methods, and significantly improves the accuracy and timeliness of abnormal event detection. The multi-level spatial mapping and virtual-reality interactive verification capabilities of dynamic digital twins enable the positioning accuracy of carbon emission sources to reach the equipment pipeline level. At the same time, through the superposition analysis of simulation data and real-time data, it provides a high-fidelity decision-making basis for process optimization.

[0018] 2. The hybrid prediction model combines the dual advantages of time series learning and physical laws. While avoiding the prediction distortion problem caused by traditional algorithms ignoring the multi-factor coupling relationship of industrial systems, it builds a model self-optimization mechanism for dynamic feedback of prediction errors to ensure that the prediction results have both data-driven flexibility and engineering physics interpretability. The multi-level early warning control system and the closed-loop tuning logic of equipment operating parameters form a complete control chain from risk warning, solution generation, virtual verification to command execution, solving the industry pain points of manual response lag and experience decision-making deviation, and realizing the transformation of carbon emission exceeding standard events from passive disposal to active prevention.

[0019] 3. At the decision support level, this application reveals the deep game relationship between carbon emissions and production efficiency through a multi-objective optimization model, and combines the virtual deduction capabilities of digital twins to provide enterprises with a Pareto optimal solution set that takes into account both economic and environmental benefits. The root cause backtracking analysis and model adversarial training mechanism further enhance the system's adaptability and fault tolerance to complex working conditions, making the technical solution transferable across industries and scales.

[0020] 4. From the perspective of extending technological value, this application not only realizes the leap of industrial carbon management from discrete monitoring to full-process intelligent control, but also builds a digital management base for carbon assets for enterprises, provides government regulatory departments with full-domain carbon flow visualization supervision tools, and promotes the innovation of collaborative carbon reduction models in the industrial chain. It has significant ecological empowerment value for the implementation of the national dual carbon strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings described herein are used to provide a further understanding of the present application and form a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation to the present application. In the drawings: Figure 1 It is a flowchart of a carbon emission monitoring and prediction method based on digital twin provided by an embodiment of the present application; Figure 2 It is a schematic diagram of the internal structure of a carbon emission monitoring and prediction device based on digital twin provided by an embodiment of the present application. Detailed implementation manners

[0022] To make the objectives, technical solutions, and advantages of the present application clearer, the technical solutions of the present application will be clearly and completely described below in conjunction with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0023] The embodiments of the present application provide a carbon emission monitoring and prediction method, device, and medium based on digital twin, which are used to implement a new carbon emission monitoring solution integrating Internet of Things perception, digital twin modeling, and intelligent decision-making.

[0024] The technical solutions proposed by the embodiments of the present application will be described in detail below with reference to the drawings.

[0025] Figure 1 It is a flowchart of a carbon emission monitoring and prediction method based on digital twin provided by an embodiment of the present application. As Figure 1 shown, a carbon emission monitoring and prediction method based on digital twin provided by an embodiment of the present application specifically includes the following steps: Step 101: Real-time collect multi-source data of the target area through multi-source heterogeneous sensors.

[0026] In this embodiment, the multi-source data includes energy flow data, material flow data, environmental flow data, equipment operation data, and spatial topology data. The multi-source heterogeneous sensor network consists of an intelligent electricity meter, an RFID reader / writer, a temperature and humidity sensor, an equipment PLC controller, and a BIM parsing module.

[0027] Exemplarily, energy flow data is collected by smart meters with a sampling frequency of not less than 100 Hz to capture instantaneous energy consumption fluctuations of devices. For example, there is a sudden increase in current during the startup phase of a rolling mill. Material flow data is tracked through RFID tags for the material transfer path and calculated in combination with the material conservation model. For example, a chemical plant scans the material RFID tags at the inlet and outlet of a reactor to calculate the matching degree of raw material consumption and product generation in real time. Environmental flow data includes parameters such as temperature, humidity, and air pressure, which are collected by environmental monitoring stations deployed in the plant area to analyze the impact of the external environment on the energy efficiency of devices. Equipment operation data is obtained through the equipment PLC controller, such as parameters like boiler combustion efficiency and compressor load rate. Spatial topology data is parsed based on the BIM model and includes the spatial coordinates of building components and equipment pipeline layout information.

[0028] It should be noted that the synchronous acquisition of multi-source data relies on timestamp alignment technology. All sensor data is embedded with a unified clock signal during acquisition to ensure that the time series of different data streams are strictly aligned. For example, when a smart meter records a sudden increase in the energy consumption of a device at a certain moment, the system synchronously retrieves the corresponding material RFID data and environmental temperature and humidity data at that moment to provide a data basis with consistent time series for subsequent correlation analysis.

[0029] Step 102: Perform dynamic anomaly detection on multi-source data, and trigger cross-data-type correlation verification when data anomalies are detected, so as to select a repair strategy according to the verification results In a possible implementation manner of the present application, performing dynamic anomaly detection on multi-source data, and triggering cross-data-type correlation verification when data anomalies are detected, so as to select a repair strategy according to the verification results specifically includes: constructing a dynamic baseline model for energy data and judging the deviation magnitude value of each data point in the energy data from the preset baseline in the dynamic baseline model; if the deviation magnitude value exceeds the preset fluctuation range, mark the corresponding data point as a suspected anomaly and synchronously verify the corresponding material conservation relationship in the material data; if the material conservation error is within the tolerance range, determine that the energy data sensor fails and trigger repair; if the material conservation error exceeds the tolerance range, determine it as a real anomaly event and trigger a multi-data stream joint repair algorithm.

[0030] In a possible implementation manner of the present application, after triggering the multi-data stream joint repair algorithm, the method further includes: extracting the temperature and humidity change characteristics in the environmental data during the abnormal period; constructing a device state transition matrix based on the start-stop records in the equipment operation log; solving the optimal repair value set that makes the energy data, material data, and environmental data compatible through a constraint satisfaction algorithm.

[0031] In this embodiment, the dynamic anomaly detection mechanism includes data baseline modeling and cross-modal verification. The data baseline model is generated by training with historical data and reflects the fluctuation range of each data stream under normal operating conditions. For example, the baseline model of the energy flow data contains the typical energy consumption curves of the equipment at different load rates, and the baseline model of the material flow data is constructed based on the law of conservation of matter to form a theoretical threshold range.

[0032] Specifically, when it is detected that a certain data stream deviates from the baseline (for example, an energy consumption value recorded by a smart meter exceeds the confidence interval of the baseline model), the system triggers a cross-data type correlation verification. For example, if at a certain moment, the energy flow data shows a sudden increase in the energy consumption of a reactor, the system synchronously verifies whether the input amount of raw materials and the output amount of products of this reactor in the material flow data match. If the material conservation error is within the tolerance range (for example, the deviation between the raw material consumption and the product generation is less than the preset threshold), it is determined that the energy flow sensor is faulty, and the data repair mechanism is triggered; if the material conservation error exceeds the limit at the same time, it is determined as a real abnormal event (for example, the energy efficiency is reduced due to a seal leak in the reactor), and the multi-data stream joint repair algorithm is triggered.

[0033] Exemplarily, when the energy consumption of a blast furnace in a steel plant is abnormal, the system discovers through correlation verification that the molten iron output does not match the coke consumption. Combining the characteristic of a sudden drop in the top temperature in the environmental flow data, it is determined as a blast furnace gas leakage event, and then the multi-data stream joint repair is triggered.

[0034] It should be noted that the joint repair algorithm solves the optimal repair value through a constraint satisfaction problem (CSP). For example, when there is a conflict between the energy flow data and the material flow data, the system extracts the environmental flow data (such as temperature and humidity changes) and equipment operation logs (such as valve opening and closing records) during the abnormal period, constructs a constraint equation set, and solves the set of data repair values that satisfy all physical laws.

[0035] Step 103: Dynamically bind the repaired data to the spatial coordinates of the BIM model to construct a three-dimensional digital twin containing the equipment-level carbon flow path.

[0036] In a possible implementation manner of the present application, dynamically binding the repaired data to the spatial coordinates of the BIM model to construct a three-dimensional digital twin containing the equipment-level carbon flow path specifically includes: in the initial mapping stage, dynamically associating the HVAC pipeline ID in the BIM model with the real-time energy consumption data; when it is detected that the carbon emissions of a specific equipment are abnormal, automatically loading the fluid dynamics simulation model of the specific equipment; overlaying and displaying the real-time sensor data and the simulation data, and characterizing the carbon emission intensity deviation through a color temperature gradient.

[0037] In this embodiment, the construction of the three-dimensional carbon flow digital twin includes two stages: spatial mapping and dynamic loading. In the spatial mapping stage, the repaired data is associated with the component IDs in the BIM model. For example, the spatial coordinates of the HVAC pipes are bound to the real-time energy consumption data, enabling the pipe models in the digital twin to dynamically display the carbon flow intensity.

[0038] Specifically, in the initial loading stage, the digital twin presents a LOD1-level simplified model (building outline and main equipment layout) and is statically associated with the basic parameters in the carbon emission factor library. When the user focuses on a specific area or an anomaly is detected, the system dynamically loads a LOD3-level detailed model (such as the topological structure of equipment pipelines) and associates real-time data streams. For example, when the carbon emission of a certain boiler is abnormal, the twin automatically loads the fluid dynamics simulation model of its combustion chamber and superimposes and displays the real-time sensor data and simulation data.

[0039] Exemplarily, in the digital twin of a data center, the carbon flow path of the cooling water pipes is visualized through color temperature gradients: blue indicates normal emissions, and red indicates excessive emissions. When a section of the pipe shows red, the user can click to view the real-time flow rate, temperature, and the operating parameters of the associated refrigeration units.

[0040] It should be noted that the dynamic loading depends on the real-time rendering optimization of the edge computing nodes. When the user zooms in or out of the view, the system dynamically adjusts the rendering accuracy according to the view range. For example, a campus-level carbon emission heat map is displayed from a macroscopic perspective, and the carbon flow details of a certain pump valve are shown from a microscopic perspective.

[0041] Step 104: Perform a rolling prediction on the carbon emission trend based on a preset hybrid prediction model, and when the predicted value triggers a multi-level early warning logic, execute control instructions from carbon emission source location to equipment parameter adjustment step by step.

[0042] In this embodiment, the hybrid prediction model consists of an LSTM time series prediction module and a physical constraint correction module. The inputs of the LSTM module include historical carbon emission sequences, production scheduling plans, weather forecast data, and equipment health indicators. For example, when predicting the carbon emissions of a certain workshop in the next 24 hours, the model considers the equipment start-stop arrangements in the production plan, the temperature change trend in the weather forecast, and the efficiency decay coefficient in the equipment maintenance records.

[0043] Specifically, the physical constraint correction module corrects the preliminary prediction results of the LSTM through thermodynamic equations and the law of conservation of mass. For example, when the LSTM predicts a sudden drop in carbon emissions during a certain period, the physical module combines the equipment thermal inertia parameters (such as the boiler cooling rate), identifies that this prediction violates the thermodynamic laws, and limits the corrected predicted value within a reasonable range.

[0044] Exemplarily, in the hybrid prediction model of a chemical plant, the LSTM predicts the carbon emission reduction rate during the cooling stage of the reactor, and the physical module smooths and corrects the prediction curve based on the thermal conductivity of the reactor material and the cooling water flow rate to avoid non-physical mutation values.

[0045] It should be noted that the model training adopts a transfer learning strategy. For example, the prediction model of a newly built park reuses the pre-trained LSTM network weights of a similar park and only fine-tunes the local data to shorten the training cycle.

[0046] In a possible implementation manner of the present application, the multi-level early warning logic includes: primary early warning logic, intermediate early warning logic, and advanced early warning logic; the primary early warning logic is to mark potential risk devices in the digital twin when the predicted value reaches the first early warning threshold; the intermediate early warning logic is to generate an optimization plan for the device operation parameters and simulate the implementation effect when the predicted value reaches the second early warning threshold; the advanced early warning logic is to trigger the device interlock protection mechanism and freeze the high-risk process section when the predicted value exceeds the third early warning threshold.

[0047] In a possible implementation manner of the present application, generating an optimization plan for the device operation parameters specifically includes: constructing a carbon emission efficiency curve for each device based on the real-time monitoring data of the three-dimensional digital twin, where the carbon emission efficiency curve reflects the carbon emissions per unit output of the device at different load rates; using the carbon emission efficiency curve as an input parameter to construct a multi-objective optimization model with the goals of minimizing carbon emissions and maximizing production efficiency; calculating the Pareto optimal solution set through the multi-objective optimization model to obtain several candidate adjustment plans including the trade-off relationship between carbon emissions and production efficiency; inputting the candidate adjustment plans into the three-dimensional digital twin for virtual execution simulation to predict the impact of each candidate adjustment plan on the carbon emission trend and the device operation stability; screening the feasible plans that meet the preset control objectives according to the simulation results, generating a device parameter priority adjustment instruction and sending it to the device control system for execution; real-time monitoring the carbon emission data after the instruction execution and sending the feedback result back to the multi-objective optimization model for dynamic weight calibration.

[0048] In this embodiment, the multi-level early warning logic is divided into primary, intermediate, and advanced early warnings. Primary early warning: When the predicted value reaches the first early warning threshold, for example, 80% of the set threshold, the system highlights potential risk devices in the digital twin and initiates diagnostic analysis. For example, when a substation predicts that the transformer load rate exceeds the standard, the twin shows the winding temperature distribution and automatically retrieves the latest maintenance records for the operation and maintenance personnel to refer to. Intermediate early warning: When the predicted value reaches the second early warning threshold, for example, 100% of the set threshold, an equipment parameter optimization plan is generated and the implementation effect is simulated. For example, after adjusting the intake valve opening of the air compressor, the twin simulates and shows that the carbon emissions are reduced and the production efficiency remains stable. After verification, the control instruction is issued. Advanced early warning: When the predicted value reaches the third early warning threshold, for example, exceeding 120% of the set threshold, the equipment interlock protection mechanism is triggered. For example, when a reactor predicts a risk of pressure out of control, the system forcibly closes the feed valve and starts the emergency cooling procedure.

[0049] It should be noted that the optimization plan is generated through a multi-objective optimization model. The model aims at minimizing carbon emissions and maximizing production efficiency, and the input parameters include the equipment carbon emission efficiency curve, real-time process constraints, and production scheduling requirements. For example, the optimization plan for a painting workshop in an automobile factory shows that reducing the drying temperature can reduce emissions by 8%, but the production cycle is extended by 2 minutes; the system recommends the comprehensive optimal plan through Pareto front analysis.

[0050] Step 105, after the control instruction is executed, verify the control effect through the real-time feedback of the three-dimensional digital twin, and initiate root cause backtracking analysis when the expected effect is not achieved.

[0051] In this embodiment, after the control instruction is executed, the system verifies the control effect through the real-time feedback of the digital twin. For example, after a boiler adjusts its combustion parameters, the twin shows the change trends of the flame shape and carbon emission intensity. If the expected effect is not achieved, root cause backtracking analysis is triggered.

[0052] In a possible implementation manner of this application, the root cause backtracking analysis specifically includes: intercepting multi-source data for each preset time period before and after the control instruction is issued; reconstructing the control instruction propagation path and equipment response time sequence in the three-dimensional digital twin; identifying the key nodes of control failure through a causal reasoning model; where the key nodes include sensor delay, model prediction deviation, or equipment actuator failure.

[0053] In this embodiment, the root cause backtracking analysis intercepts the multi-source data streams before and after the control instruction is issued and reconstructs the event space-time path. For example, during a control failure, the system locates a communication module failure of a certain PLC controller by analyzing the equipment response delay data in the digital twin and generates a maintenance work order.

[0054] Exemplarily, in the root cause tracing of a certain data center, the system found that the carbon emissions did not decrease after the air conditioning system was adjusted. By analyzing the airflow simulation data in the digital twin, a design defect was identified in a certain air supply pipeline, and pipeline renovation suggestions were put forward.

[0055] It should be noted that the root cause analysis relies on a causal reasoning model. The model constructs a causal relationship graph of equipment parameters, environmental variables, and carbon emission results through a Bayesian network to identify key fault nodes. For example, a certain prediction deviation was traced back to the drift of the humidity sensor, and the system then triggered the sensor calibration program.

[0056] In a possible implementation manner of the present application, the method further includes: when the root cause tracing analysis identifies a prediction model deviation, automatically extracting the feature vector of the failure case; injecting the feature vector into the adversarial training set to generate a new model version with perturbation robustness; running the old and new models in parallel in the three-dimensional digital twin, and selecting the optimal model to go online through real-time data verification.

[0057] In this embodiment, when the root cause analysis identifies a model prediction deviation, the system automatically extracts the feature vector of the failure case and injects it into the adversarial training set to generate a new robust model. For example, a certain LSTM model's prediction was inaccurate due to not considering the impact of extreme weather. After adding high-temperature scenario data to the adversarial training set, the prediction error of the new model under similar working conditions was significantly reduced.

[0058] Specifically, the old and new models run in parallel in the digital twin, and the optimal version is selected through real-time data verification. For example, a certain park switches to a humidity enhancement model during the plum rain season, and the digital twin synchronously displays the comparison of the prediction curves of the old and new models. After the operation and maintenance personnel confirm that the new model has higher accuracy, the online switch is completed.

[0059] It should be noted that the dynamic weight calibration mechanism adjusts the model parameters in real time according to the control feedback data. For example, when the energy efficiency of a certain device decreases due to aging, the system automatically increases the training weight of its historical data to make the prediction model adapt to the change of the device state.

[0060] The above is the method embodiment proposed in the present application. Based on the same inventive concept, the embodiment of the present application also provides a digital twin-based carbon emission monitoring and prediction device, the structure of which is as Figure 2 shown.

[0061] Figure 2 This is a schematic diagram of the internal structure of a digital twin-based carbon emission monitoring and prediction device provided by the embodiment of the present application. As Figure 2 shown, the device includes: At least one processor 201; And a memory 202 communicatively connected to at least one processor; Among them, the memory 202 stores instructions executable by at least one processor, and the instructions are executed by at least one processor 201 so that at least one processor 201 can: Collect multi-source data of the target area in real time through multi-source heterogeneous sensors; among them, the multi-source data includes: energy data, material data, environmental data, equipment operation data, and spatial topology data; Perform dynamic anomaly detection on the multi-source data, and trigger cross-data-type correlation verification when data anomalies are detected, so as to select a repair strategy according to the verification results; Dynamically bind the repaired data to the spatial coordinates of the BIM model to construct a three-dimensional digital twin containing the device-level carbon flow path; Based on a preset hybrid prediction model, perform rolling prediction on the carbon emission trend, and when the predicted value triggers a multi-level early warning logic, gradually execute control instructions from carbon emission source location to equipment parameter adjustment; After the control instructions are executed, verify the control effect through the real-time feedback of the three-dimensional digital twin, and start root cause backtracking analysis when the expected effect is not achieved.

[0062] Some embodiments of the present application provide a non-volatile computer storage medium corresponding to Figure 1 for carbon emission monitoring and prediction based on digital twin, storing computer-executable instructions, and the computer-executable instructions are set as: Collect multi-source data of the target area in real time through multi-source heterogeneous sensors; among them, the multi-source data includes: energy data, material data, environmental data, equipment operation data, and spatial topology data; Perform dynamic anomaly detection on the multi-source data, and trigger cross-data-type correlation verification when data anomalies are detected, so as to select a repair strategy according to the verification results; Dynamically bind the repaired data to the spatial coordinates of the BIM model to construct a three-dimensional digital twin containing the device-level carbon flow path; Based on a preset hybrid prediction model, perform rolling prediction on the carbon emission trend, and when the predicted value triggers a multi-level early warning logic, gradually execute control instructions from carbon emission source location to equipment parameter adjustment; After the control instructions are executed, verify the control effect through the real-time feedback of the three-dimensional digital twin, and start root cause backtracking analysis when the expected effect is not achieved.

[0063] The various embodiments in the present application are all described in a progressive manner. The same or similar parts among the various embodiments can be referred to each other, and the key points of each embodiment are the differences from other embodiments. In particular, for the embodiments of the Internet of Things devices and media, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0064] The system, medium, and method provided by the embodiments of the present application correspond one by one. Therefore, the system and the medium also have beneficial technical effects similar to those of the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the system and the medium will not be elaborated here.

[0065] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0066] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the flows and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one or more of the processes or Figure 1 blocks.

[0067] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implements the functions specified in Figure 1 one or more of the processes or Figure 1 blocks.

[0068] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one or more of the processes or Figure 1 blocks.

[0069] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0070] Memory may include non-permanent storage in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0071] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0072] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0073] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A carbon emission monitoring and prediction method based on digital twins, characterized in that: The method comprises: Collect multi-source data of the target area in real time through multi-source heterogeneous sensors; wherein the multi-source data includes: energy data, material data, environmental data, equipment operation data and spatial topology data; Performing dynamic anomaly detection on the multi-source data, and triggering correlation verification across data types when data anomalies are detected, so as to select a repair strategy according to the verification results; Dynamically bind the repaired data to the spatial coordinates of the BIM model to build a three-dimensional digital twin containing the equipment-level carbon flow path; Carry out rolling forecasts of carbon emission trends based on the preset hybrid forecasting model, and when the forecast value triggers the multi-level early warning logic, execute control instructions from carbon emission source location to equipment parameter adjustment step by step; After the control instruction is executed, the control effect is verified through the real-time feedback of the three-dimensional digital twin, and the root cause backtracking analysis is initiated when it does not meet expectations.

2. According to a carbon emission monitoring and prediction method based on digital twins according to claim 1, it is characterized in that: Dynamic anomaly detection is performed on the multi-source data, and when data anomalies are detected, correlation verification across data types is triggered to select a repair strategy based on the verification results, specifically including: Constructing a dynamic baseline model for the energy data, and determining the deviation value of each data point in the energy data from a preset baseline in the dynamic baseline model; If the deviation value exceeds the preset fluctuation range, the corresponding data point is marked as suspected abnormality, and the corresponding material conservation relationship in the material data is verified simultaneously; If the material conservation error is within the tolerance range, the energy data sensor is judged to be faulty and repair is triggered; If the material conservation error exceeds the tolerance range, it is determined to be a real abnormal event, triggering the multi-data stream joint repair algorithm.

3. The carbon emission monitoring and prediction method based on digital twin according to claim 2 is characterized in that: After triggering the multi-data stream joint repair algorithm, the method further includes: Extract the temperature and humidity change characteristics from the environmental data during the abnormal period; Construct the equipment state transition matrix based on the start and stop records in the equipment operation log; The constraint satisfaction algorithm is used to solve the optimal set of repair values ​​that make energy data, material data, and environmental data compatible.

4. The carbon emission monitoring and prediction method based on digital twin according to claim 1 is characterized in that: Dynamically bind the repaired data to the spatial coordinates of the BIM model to build a three-dimensional digital twin containing the equipment-level carbon flow path, including: In the initial mapping stage, the HVAC pipe ID in the BIM model is dynamically associated with the real-time energy consumption data; When abnormal carbon emissions from a specific device are detected, a fluid dynamics simulation model of the specific device is automatically loaded; The real-time sensor data is superimposed with the simulation data, and the carbon emission intensity deviation is represented by the color temperature gradient.

5. The carbon emission monitoring and prediction method based on digital twin according to claim 1 is characterized in that: The multi-level warning logic includes: primary warning logic, intermediate warning logic, and advanced warning logic; The primary warning logic is to mark the potential risk equipment in the digital twin when the predicted value reaches the first warning threshold; The intermediate warning logic is to generate an equipment operation parameter optimization plan and simulate the implementation effect when the predicted value reaches the second warning threshold; The advanced warning logic is that when the predicted value exceeds the third warning threshold, the equipment interlock protection mechanism is triggered and the high-risk process section is frozen.

6. The carbon emission monitoring and prediction method based on digital twin according to claim 5 is characterized in that: Generate equipment operation parameter optimization plan, including: Based on the real-time monitoring data of the three-dimensional digital twin, a carbon emission efficiency curve of each device is constructed, wherein the carbon emission efficiency curve reflects the carbon emission per unit output of the device at different load rates; Taking the carbon emission efficiency curve as an input parameter, a multi-objective optimization model with the goals of minimizing carbon emissions and maximizing production efficiency is constructed; Calculating the Pareto optimal solution set through the multi-objective optimization model to obtain several candidate adjustment plans that include a trade-off relationship between carbon emissions and production efficiency; Inputting the candidate adjustment schemes into the three-dimensional digital twin for virtual execution simulation to predict the impact of each candidate adjustment scheme on carbon emission trends and equipment operation stability; Screen feasible solutions that meet the preset control objectives based on simulation results, generate equipment parameter priority adjustment instructions and send them to the equipment control system for execution; The carbon emission data after the instruction is executed is monitored in real time, and the feedback results are sent back to the multi-objective optimization model for dynamic weight calibration.

7. The carbon emission monitoring and prediction method based on digital twin according to claim 1 is characterized in that: Root cause retrospective analysis, including: Intercept multi-source data in preset time periods before and after the control command is issued; reconstructing a control instruction propagation path and a device response timing in the three-dimensional digital twin; The key nodes of control failure are identified through a causal reasoning model; wherein the key nodes include sensor delays, model prediction deviations, or device actuator failures.

8. The carbon emission monitoring and prediction method based on digital twin according to claim 7 is characterized in that: The method further comprises: Automatically extract feature vectors of failure cases when root cause backtracking analysis identifies deviations from the prediction model; Inject the feature vector into the adversarial training set to generate a new model version that is robust to perturbations; The new and old models are run in parallel in the three-dimensional digital twin, and the optimal model is selected for online launch through real-time data verification.

9. A carbon emission monitoring and prediction device based on digital twins, characterized in that: The device comprises: at least one processor; and, a memory communicatively coupled to the at least one processor; In which, the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a carbon emission monitoring and prediction method based on digital twins as described in any one of claims 1-8.

10. A non-volatile computer storage medium for carbon emission monitoring and prediction based on digital twins, storing computer executable instructions, characterized in that: When the computer executable instructions are executed, a carbon emission monitoring and prediction method based on digital twins as described in any one of claims 1 to 8 is implemented.

Images